2018 in Review
This article looks at my role as a Statistical Ambassador for the Royal Statistical Society in 2018, as well as updating my analysis of Doctor Who viewing figures.
Kerblam!
First, I will update some previous analysis on Doctor Who viewing figures.
We can now add 7-day consolidated viewer estimates from BARB, for the final four episodes of the series.

There was a large reduction in viewers after the first episode, followed by a general decline:

Series 11 received an estimated average of 7.96m viewers per episode, making it the second-most watched series of Doctor Who since the revival.
This series was shorter than previous iterations, standing at only 10 episodes. The final episode did relatively poorly versus the last episodes from previous series. The Battle of Ranskoor Av Kolos was estimated to have been watched by 6.65m viewers in the seven days since its broadcast. This ranks eighth of out 11: beating only The Big Bang (6.12m), Hell Bent (6.17m) and The Doctor Falls (5.29m).
We can now show the whole series as a index of its average episode, showing the broad erosion of viewers over the course of a series:

Whether you create an index based on the first episode or the series average, Series 11 had the lowest viewer index for its final episode of all revived series (60.7% of its first episode, 83.6% of the series average).
This is reflective of relatively high viewing figures in the early episodes declining to average viewer levels by the later episodes.
The data-set is available online, and I used an R console to create the graphs.
2018 in Review
In the first half of this year, I was selected to be a Statistical Ambassador for the Royal Statistical Society. I was part of the third cohort of Ambassadors. This role is voluntary, seeking to engage and inform journalists, policy-makers, researchers and the wider public about statistics.
We undertook media training, learning how to cope with being on radio and television, and how to talk clearly about difficult statistical concepts. We also had professional photos taken by Big T Images.

Since June, I have:
- Engaged with a market research company and the Market Research Society regarding a strange survey result, publicly publishing their data tables;
- Helped the BBC by calculating the probability of winning the US Mega Millions jackpot with a single ticket, and provided contextual probabilities;
- Written an original article for The Spectator, pointing out flaws in the estimation it was “very likely” that Vote Leave’s overspending caused 800,000 voters to switch their choice in the 2016 EU referendum;
- Fact-checked submissions for the Royal Statistical Society’s Statistic of the Year panel.
There have been smaller successes and some failures too, which I have not written about before.
In October, I contacted the BBC about a graph contained in an article on car registration data. The bar graph had previously started its axis at 320,000, which is disproportionate. This was corrected:

I contacted ITV regarding their claim that “52% of Brits think the Moon landings from 1969–72 were faked”, pointing out differences between that result and other known surveys. I received no substantive reply.
I asked for a headline to be corrected in The Independent about the NatCen mixed-mode survey. It is based on a panel, which had become increasingly Remain-heavy over time — meaning the ‘raw’ result is less meaningful. The journalist quickly did so, for which I am grateful.
I also contacted the Daily Star, iNews and RTÉ News about similar headlines. An unsuccessful complaint to IPSO meant these were not corrected.
The Year Ahead
I have met many wonderful people through the Statistical Ambassador scheme.
I look forward to the year ahead, when there will be lots more statistics and probabilities to be better understood.